I'm troubleshooting an intermittent performance problem on my gaming PC. The system has a Ryzen 5 7500F, an RTX 5070 12GB, 16GB of DDR5-5600 memory, and a 750W power supply. In any game, the FPS occasionally drops sharply after roughly one or two hours, causing obvious stuttering. After a while, performance returns to normal.
MSI Afterburner shows that either CPU or GPU usage can fall suddenly during the slowdown, but they do not always drop together. Temperatures appear normal at around 65°C, system memory usage is usually about 12GB out of 16GB, and VRAM usage is around 6GB out of 12GB. The CPU and GPU drivers were updated about a month ago, and several other troubleshooting steps have not resolved the issue.
What could cause this kind of intermittent drop in utilization and FPS? Should I investigate power limits, clock speeds, motherboard VRM or chipset temperatures, GPU hotspot temperature, storage temperatures, background software, Windows, or something else?
3 Answers
Since the issue happens across different games and the memory does not appear to be full, look for a system-wide trigger such as background software, scheduled tasks, overlays, or a power-plan change. Keep logging CPU/GPU clocks, board power, temperatures, and frametimes so you can see exactly what changes during the stutter. GPU hotspot temperature and clock speed are especially important because the reported core temperature can still look fine while the card throttles.
Check the actual power draw and clock speeds of both the CPU and GPU at the moment performance drops. A sudden reduction in power or clocks could point to throttling, a power-management state, or a driver issue. Monitoring only overall temperature and utilization may miss the real cause.
It would be worth checking temperatures beyond the basic CPU and GPU readings. Monitor the GPU hotspot, motherboard VRM or MOSFET sensors, chipset, and NVMe drive temperature while gaming. A hotspot or motherboard component can briefly throttle even when the main CPU and GPU temperatures look completely normal. Running separate CPU and GPU stress tests with OCCT could help reproduce the problem.
The motherboard is a B650E model with heatsinks over the power circuitry. I’ll run OCCT tests and monitor the additional temperature sensors rather than relying only on the main CPU and GPU readings.